During a press conference prior to the release of the executive summary of 3-year study of trends and future projections for the factors and determinants that underpin malaria by its Strategic Advisory Group on Malaria Eradication (SAGme), WHO outlined some hopeful signs emanating from the SAGme including
Lack of biological barriers to malaria eradication
Recognition of the massive social and economic benefits that would provide a return on investment in eradication, and
Megatrends in the areas of factors such as land use, climate, migration, urbanization that could inhibit malaria transmission
Concerning the first point, the executive summary notes that, “We did not identify biological or environmental barriers to malaria eradication. In addition, our review of models accounting for a variety of global trends in the human and biophysical environment over the next three decades suggest that the world of the future will have much less malaria to contend with.”
The group did agree that, “using current tools, we will still have 11 million cases of malaria in Africa in 2050.” So one wonders whether there are biological barriers or not.
Interestingly the group did identify, “Potential biological threats to malaria eradication include development of insecticide and antimalarial drug resistance, vector population dynamics and altered vector behaviour. For example, Anopheles vectors might adapt to breeding in polluted water, and mosquito vector species newly introduced to Africa, such as Anopheles stephensi, could spread more widely into urban settings.”
Their further expansion on the biological issues using smallpox as an example is instructive. They noted that not only are humans essential for the life cycle of the organism, but that there was no other reservoir for the causative virus, and the virus could not amplify in the environment. In short, there were no vectors, as in the case of malaria. The relatively recent documentation of transmission of malaria between humans and other primates of different plasmodium species is another biological concern. At this point, Malaysia, for example, is reporting more cases of Plasmodium knowlesi in humans that either P vivax or P falciparum.
Another biological issue identified by Aylward and colleagues was the fact that smallpox had one effective and proven intervention, the vaccine. Application of the vaccine could be targeted using photograph disease recognition cards as the signs were quite specific to the disease. Malaria has several effective interventions, but most strategies emphasize the importance of using a combination of these, and implementation is met with a number of management and logistical challenges. The signs and symptoms of malaria are confused with a number of febrile illnesses.
Finally, two other issues raised concern. Insecticide resistance was recognized in the first malaria eradication effort, and is raising its head again, as pointed out by SAGme. Comparing smallpox and yaws, the challenge of latent or sub-clinical/asymptomatic infection was mentioned. Malaria too, is beleaguered with this problem.
Clearly, we must not lose momentum in the marathon (not a race) to eliminate malaria, but we must, as WHO stressed at the press conference, increase our research and development efforts to strengthen existing tools and develop new once to address the biological and logistical challenges.
If all it took to eradicate a disease was a well proven drug, vaccine or technology, we would not be still reporting on polio, measles and guinea worm, to name a few. In the past week Afghanistan reported 2 wild poliovirus type 1 (WPV1) cases, and Pakistan had 3 WPV1 cases. Circulating vaccine-derived poliovirus type 2 (cVDPV2) was reported in Nigeria (1), DRC (4) and Ethiopia (3) from healthy community contacts.
Pacific Standard explained the differences in Ebola outbreaks between DRC today and the West Africa outbreak of 2014-16. On the positive side are new drugs used in organized trials for the current outbreak. The most important factor is safe, effective vaccine that has been tested in 2014-16, but is now a standard intervention in the DRC. While both Liberia and Sierra Leone had health systems and political weaknesses as post-conflict countries, DRC’s North Kivu and Ituri provinces are currently a war zone, effectively so for the past generation. Ebola treatment centers and response teams are being attacked. There are even cultural complications, a refusal to believe that Ebola exists. So even with widespread availability of improved technologies, teams may not be able to reach those in need.
To further complicate matters in the DRC, Doctors Without Borders (MSF) “highlighted ‘unprecedented’ multiple crises in the outbreak region in northeastern DRC. Ebola is coursing through a region that is also seeing the forced migration of thousands of people fleeing regional violence and is dealing with another epidemic. Moussa Ousman, MSF head of mission in the DRC, said, ‘This time we are seeing not only mass displacement due to violence but also a rapidly spreading measles outbreak and an Ebola epidemic that shows no signs of slowing down, all at the same time.’”
NIPAH and Bats
Like Ebola, NIPAH is zoonotic, and also involves bats, but the viruses differ. CDC explains that, “Nipah virus (NiV) is a member of the family Paramyxoviridae, genus Henipavirus. NiV was initially isolated and identified in 1999 during an outbreak of encephalitis and respiratory illness among pig farmers and people with close contact with pigs in Malaysia and Singapore. Its name originated from Sungai Nipah, a village in the Malaysian Peninsula where pig farmers became ill with encephalitis.
A recent human outbreak in southern India has been followed up with a study of local bats. In a report shared by ProMED, out of 36 Pteropus species bats tested for Nipah, 12 (33%) were found to be positive for anti-Nipah bat IgG antibodies. Unlike Ebola there are currently no experimental drugs or vaccines.
Climate Change and Dengue
Climate change is expected to heighten the threat of many neglected tropical diseases, especially arboviral infections. For example, the New York Times reports that increases in the geographical spread of dengue fever. Annually “there are 100 million cases of dengue infections severe enough to cause symptoms, which may include fever, debilitating joint pain and internal bleeding,” and an estimated 10,000 deaths. Dengue is transmitted by Aedes mosquitoes that also spread Zika and chikungunya. A study, published Monday in the journal Nature Microbiology, found that in a warming world there is a strong likelihood for significant expansion of dengue in the southeastern United States, coastal areas of China and Japan, as well as to inland regions of Australia. “Globally, the study estimated that more than two billion additional people could be at risk for dengue in 2080 compared with 2015 under a warming scenario.”
Schistosomiasis – MDA Is Not Enough, and Neither Are Supplementary Interventions
Schistosomiasis is one of the five neglected tropical diseases (NTDs) that are being controlled and potentially eliminated through mass drug administration (MDA) of preventive chemotherapy (PCT), in this case praziquantel. In The Lancet Knopp et al. reported that biannual MDA substantially reduced Schistosomiasis haematobium prevalence and infection intensity but was insufficient to interrupt transmission in Zanzibar. In addition, neither supplementary snail control or behaviour change activities did not significantly boost the effect of MDA. Most MDA programs focus on school aged children, and so other groups in the community who have regular water contact would not be reached. Water and sanitation activities also have limitations. This raises the question about whether control is acceptable for public health, or if there needs to be a broader intervention to reach elimination?
Trachoma on the Way to Elimination
Speaking of elimination, WHO has announced major “sustained progress” on trachoma efforts. “The number of people at risk of trachoma – the world’s leading infectious cause of blindness – has fallen from 1.5 billion in 2002 to just over 142 million in 2019, a reduction of 91%.” Trachoma is another NTD that uses the MDA strategy.
The news about NTDs from Dengue to Schistosomiasis to Trachoma is complicated and demonstrates that putting diseases together in a category does not result in an easy choice of strategies. Do we control or eliminate or simply manage illness? Can our health systems handle the needs for disease elimination? Is the public ready to get on board?
And concerning being complicated, malaria this week again shows many facets of challenges ranging from how to recognize and deal with asymptomatic infection to preventing reintroduction of the disease once elimination has been achieved. Several reports this week showed the particular needs for malaria intervention ranging from high burden areas to low transmission verging on elimination to preventing re-introduction in areas declared free from the disease.
In South West, Nigeria Dokunmu et al. studied 535 individuals aged from 6 months were screened during the epidemiological survey evaluating asymptomatic transmission. Parasite prevalence was determined by histidine-rich protein II rapid detection kit (RDT) in healthy individuals. They found that, “malaria parasites were detected by RDT in 204 (38.1%) individuals. Asymptomatic infection was detected in 117 (57.3%) and symptomatic malaria confirmed in 87 individuals (42.6%).
Overall, detectable malaria by RDT was significantly higher in individuals with symptoms (87 of 197/44.2%), than asymptomatic persons (117 of 338/34.6%)., p = 0.02. In a sub-set of 75 isolates, 18(24%) and 14 (18.6%) individuals had Pfmdr1 86Y and 1246Y mutations. Presence of mutations on Pfmdr1 did not differ by group. It would be useful for future study to look at the effect of interventions such as bednet coverage. While Southwest Nigeria is a high burden area, the problem of asymptomatic malaria will become an even bigger challenge as prevalence reduces and elimination is in sight.
Sri Lanka provides a completely different challenge from high burden areas. There has been no local transmission of malaria in Sri Lanka for 6 years following elimination of the disease in 2012. Karunasena et al. report the first case of introduced vivax malaria in the country by diagnosing malaria based on microscopy and rapid diagnostic tests. “The imported vivax malaria case was detected in a foreign migrant followed by a Plasmodium vivax infection in a Sri Lankan national who visited the residence of the former. The link between the two cases was established by tracing the occurrence of events and by demonstrating genetic identity between the parasite isolates. Effective surveillance was conducted, and a prompt response was mounted by the Anti Malaria Campaign. No further transmission occurred as a result.”
Bangladesh has few but focused areas of malaria transmission and hopes to achieve elimination of local transmission by 2030. A particular group for targeting interventions is the population of slash and burn cultivators in the Rangamati District. Respondents in this area had general knowledge about malaria transmission and modes of prevention and treatment was good according to Saha and the other authors. “However, there were some gaps regarding knowledge about specific aspects of malaria transmission and in particular about the increased risk associated with their occupation. Despite a much-reduced incidence of malaria in the study area, the respondents perceived the disease as life-threatening and knew that it needs rapid attention from a health worker. Moreover, the specific services offered by the local community health workers for malaria diagnosis and treatment were highly appreciated. Finally, the use of insecticide-treated mosquito nets (ITN) was considered as important and this intervention was uniformly stated as the main malaria prevention method.”
Kenya offers some lessons about low transmission areas but also areas where transmission may increase due to climate change. A matched case–control study undertaken in the Western Kenya highlands. Essendi et al. recruited clinical malaria cases from health facilities and matched to asymptomatic individuals from the community who served as controls in order to identify epidemiological risk factors for clinical malaria infection in the highlands of Western Kenya.
“A greater percentage of people in the control group without malaria (64.6%) used insecticide-treated bed nets (ITNs) compared to the families of malaria cases (48.3%). Low income was the most important factor associated with higher malaria infections (adj. OR 4.70). Houses with open eaves was an important malaria risk factor (adj OR 1.72).” Other socio-demographic factors were examined. The authors stress the need to use local malaria epidemiology to more effectively targeted use of malaria control measures.
The key lesson arising from the forgoing studies and news is that disease control needs strong global partnerships but also local community investment and adaptation of strategies to community characteristics and culture.
After the World’s first attempt at eradicating the
complicated disease malaria mainly through a single tool, a period of control
set in where the aim was to reduce mortality through prompt and presumptive
treatment of fevers with anti-malarials, particularly in young children. During
this period in the 1980s and 1990s it was recognized that parasite-based
diagnostic capabilities in the form of microscopy were limited, so in malaria
endemic areas, it was worth providing inexpensive medicines like chloroquine
(CQ) and sulfadoxine-pyrimethamine (SP) to febrile children in order to save lives.
When the fevers did not resolve, other illnesses explored.
The difficulty arose in identifying cases that did not offer
clinical clues that they might be malaria. Today countries approaching malaria
elimination face challenges, such as seen in Zanzibar where, “outdoor
transmission, a large asymptomatic parasite reservoir and imported infections,
require novel tools and reoriented strategies to prevent a rebound effect and
Here we examine the challenge of asymptomatic malaria infections.
By 1998 when the Roll Back Malaria partnership formed, there
had been enough research done so that the malaria community had a better
arsenal of interventions including insecticide-treated bed nets,
artemisinin-based combination therapy (ACT) and intermittent preventive
treatment with SP during pregnancy. The Abuja Declaration of 2000 set a target
of 80% coverage of these interventions by the year 2010.
While ACTs overcame the challenges of parasite resistance
that had developed for the single drugs, CQ and SP, it cost several times more
than those medicines. The need for easy-to-use, inexpensive, point-of-care
diagnostics was recognized so that not only would ACTs be targeted only to
parasitologically confirmed malaria cases, but also in the process, overuse and
misuse would not contribute to parasite resistance of these new drugs.[ii]
Unfortunately, the development and dissemination of antigen-based rapid
diagnostic tests (RDTs), lagged behind the availability of ACTs meaning that
health workers unfortunately continued their business as usual with presumptive
treatment using ACTs.
The benefits of RDTs were generally two-fold. First, they
could be used by front-line, auxiliary and community-based health workers.
Secondly, they tended to identify more cases than microscopy. The big challenge
was convincing health workers to use them and trust the results, because the
era of presumptive treatment had given these staff a false sense of confidence
in their own clinical diagnostic abilities.
Although reaching the 2010 coverage targets has remained
illusive for most endemic countries, there has been enough progress for major
reductions in incidence (despite a recent upsurge).[iii]
As the proportion of actual malaria cases among febrile illness patients
declines, concern has risen that transmission might continue among people with
subclinical or asymptomatic malaria. Here we explore the extent of this problem
and new directions in parasitological testing needed to ensure continued
progress toward elimination in each endemic country.
Understanding the Risk of Asymptomatic Malaria
Risk can relate to geographical, epidemiological, and socio-demographic factors as well as history of malaria interventions. Kenya has stratified the country by higher and lower malaria transmission areas. Even the higher areas are comparatively low compared to its higher transmission neighbors. Studying the prevalence of asymptomatic malaria in some of these higher transmission areas in the west of the country was seen as a way to better identify people at risk and learn about intervention effectiveness. An examination of apparently healthy children (no symptoms) revealed a Plasmodium falciparum malaria prevalence 36.0% (27.5%, 44.5%) by RDT and 22.3% (16.0%, 28.6%) by thick film microscopy.[iv] Living in a household with electricity was protective but the adjusted odds ratio of prevalence comparing households with and without indoor residual spray showed only borderline benefit. Unfortunately, in Zanzibar, asymptomatic malaria infection was not associated “with use of any vector control.”1
A major challenge in detecting cases through routine health
care systems is care seeking patterns of care seeking for fever. The 2018 World
Malaria Report acknowledges that there are major equity challenges in care
seeking wherein families with higher incomes, better education and living in
urban areas are more likely to seek help for their febrile children that rural,
poor and less educated families who would be more at risk. Care seeking without
the signs of fever is more challenging. A dual strategy of enabling better
service utilization as well as outreach to detect cases will be necessary to
detect asymptomatic cases.3
In Burkina Faso, the prevalence of asymptomatic malaria
infection in children under 5 years of age was estimated at 38.2% in 24 of its
70 health districts. Those at most risk for asymptomatic malaria infection
included the following:[v]
older children (48–59 vs < 6 months: OR: 6.79
children from very poor households (Richest vs
poorest: OR: 0.85 [0.74–0.96])
households located more than 5 km from a health
facility (< 5 km vs ? 5 km: OR: 1.14 [1.04–1.25])
localities with inadequate number of nurses
(< 3 vs ? 3: 0.72 [0.62, 0.82]
rural areas (OR: 1.67 [1.39–2.01])
Nine districts reported significantly higher risks (Batié,
Boromo, Dano, Diébougou, Gaoua, Ouahigouya, Ouargaye, Sapouy and Toma. The
researchers concluded that, “Such national spatial analysis should help to
prioritize areas for increased malaria control activities.”
A study in Ghana found that, “children and pregnant women had higher prevalence of submicroscopic gametocytes (39.5% and 29.7%, respectively) compared to adults (17.4%).”[vi]
An additional concern is emerging in terms of sharing of malaria parasite species between humans and primates, especially as urbanization and deforestation push these two populations into closer contact. For example Mapua and colleagues working in Central Africa Republic, “found the human malaria parasite P. ovale wallikeri in both asymptomatic humans and western lowland gorillas in Dzanga Sangha Protected Areas. Molecular analysis revealed that the genotype of the P. ovale wallikeri DNA found in a gorilla was genetically identical to that of a human isolate within the mt cytb and mt cox 1 genes, indicating potential human–ape transmission.”[vii] They noted similar sharing of parasites in the region between humans and chimpanzees.
Detecting and Responding to Asymptomatic Cases
WHO’s Framework for Malaria Elimination[viii]
recognizes the important role of case detection and subsequent treatment as
well as broader community level preventive responses around detected cases. In
the context of elimination WHO notes that case detection “requires use of
a diagnostic test to identify asymptomatic malaria infections.” WHO
stresses that a case is a case, regardless of whether it is symptomatic or
asymptomatic, as long as the diagnostic process confirms presence of malaria
It is important to monitor Plasmodium parasitemia in areas where malaria
transmission has declined and efforts to achieve malaria elimination are
underway, such as Zambia, where 3,863 household members were tested.[ix]
Only 2.6% were positive by either microscopy, RDT, or PCR. Of these, 48 (47%)
had subpatent parasitemia, and 85% of those with subpatent parasitemia were
asymptomatic. “Compared with individuals without parasitemia, individuals with
subpatent parasitemia were significantly more likely to be aged 5–25 years.”
The authors suggested that their findings pointed to the need for active or
reactive case detection to identify asymptomatic individuals and thus better
target individuals with subpatent parasitemia with appropriate malaria
WHO explains that active case detection (ACD) takes place in
areas of limited or under-utilization of health care services.4 It
may start with initial screening for symptoms, followed by appropriate
parasitological laboratory confirmation. In low-transmission settings or as
part of a focus investigation, “ACD may consist of testing of a defined
population group without prior symptom screening (population-wide or mass
testing) in order to identify asymptomatic infections.” Elimination cannot be
achieved until even asymptomatic infections have stopped. The challenge is the
expense of community-wide screening.
Reactive Case Detection (RCD), according to WHO, takes place
in settings low transmission intensity where the few “occurring malaria cases
are highly aggregated.”4 When a case is identified, usually through
identification of an actual infected patient at a local clinic, the community
where the patient comes from is visited and a “net is cast around the
index case” where household members and neighbors within a selected radius
are tested. In this process asymptomatic cases are also identified.
Our existing diagnostic tools may be inadequate. McCreesh
and colleagues reported on subpatent malaria in Namibia that, “fever
history and standard RDTs are not useful to address this burden. Achievement of
malaria elimination may require active case detection using more sensitive
point-of-care diagnostics or presumptive treatment and targeted to high-risk
groups.” This includes loop-mediated isothermal amplification (LAMP) using
dried blood spots, which they tested.[x]
Likewise from experience in a Zambian study, Kobayashi and co-researchers
suggest, “more sensitive diagnostic tests or focal drug administration may be
necessary to target individuals with subpatent parasitemia to achieve malaria elimination.”[xi]
Responses to detecting asymptomatic cases start at the
individual level with prompt treatment of those found through RCD to be
infected. Then focused preventive interventions such as distribution of
insecticide treated bednets can be provided to those in the cluster or village.
Follow-up would be needed for such ‘hot spots.’
On a broader basis we have Seasonal Malaria Chemoprevention
(SMC) as practiced in Sahelian countries where during the peak transmission
(rainy) season intermittent preventive treatment is given to children monthly
by community health workers and volunteers. Of course, many of these children
would be asymptomatic carriers and SMC could benefit the reduction of parasites
in circulation. At present SMC focuses on pre-school aged children, but Thera
and co-researchers stress the importance of reaching school aged children who
are also often asymptomatic carriers.[xii]
Another intervention being tested for mass drug
administration (MDA) use providing the community with ivermectin, a drug that
has been highly effective in controlling filarial diseases and also found to
kill mosquitoes who take a blood meal from a person who has recently taken it.[xiii]
This strategy is still being tested, but again MDA means all community members,
especially those with asymptomatic infection, would be reached.
A major question requires further research. To what extent
do asymptomatic, submicroscopic and subpatent parasitemia contribute to
continued malaria transmission? Another question is how can we address malaria
infection in other primates? We know that scientists recommend targeting of
malaria elimination interventions based on mapping of these infections.5
We therefore need to study the actual transmission potential of this
A, Shakely D, Ali AS, Morris U, Mkali H, Abbas AK, Al-Mafazy A-W, Haji KA, Mcha
J, Omar R, Cook J, Elfving K, Petzold M, Sachs MC, Aydin-Schmidt B, Drakeley V,
Msellem M and Mårtensson A. From high to low malaria transmission in
Zanzibar—challenges and opportunities to achieve elimination. BMC Medicine
(2019) 17:14, https://doi.org/10.1186/s12916-018-1243-z
Malaria Programme. Universal access to malaria diagnostic testing – An operational
manual. World Health Organization. November 2011 (rev. February 2013).
Malaria Programme. World malaria report 2018. World Health Organization. 19
November 2018. https://www.who.int/malaria/publications/world-malaria-report-2018/en/
S, Tenge C, Genga IO, Mumia M, Were PA, Kuremu RT, Wekes WN, Sumba PO, Kinyera T, Otim T, Legason ID,
Biddle J, Reynolds SJ, Talisuna AO, Biggar1 RJ, Bhatia K, Goedert JJ, Pfeiffer
RM, Mbulaiteye SM. A Cross-Sectional Population Study of Geographic,
Age-Specific, and Household Risk Factors for Asymptomatic Plasmodium falciparum
Malaria Infection in Western Kenya. The American Journal of Tropical Medicine
and Hygiene, Volume 100, Issue 1, Jan 2019, p.54-65. DOI:
Ouédraogo M, Samadoulougou S, Rouamba T, Hien H, Sawadogo JEM Tinto H, Alegana
VA, Speybroeck N and Kirakoya?Samadoulougou F. Spatial distribution and
determinants of asymptomatic malaria risk among children under 5 years in 24
districts in Burkina Faso. Malaria Journal 2018; 17:460
H, Ofori MF, Kusi KA, Adu B, Owusu-Yeboa E, Kyei-Baafour E, Arku AT, Bosomprah
S, Alifrangis M, Quakyi IA. The prevalence of submicroscopic Plasmodium
falciparum gametocyte carriage and multiplicity of infection in children,
pregnant women and adults in a low malaria transmission area in Southern Ghana.
Malar J. 2018 Sep 17;17(1):331. doi: 10.1186/s12936-018-2479-y.
MI, Hans-Peter Fuehrer HP, Petrželková KJ, Todd A, Noedl H, Qablan MA, and
Modrý D. Plasmodium ovale wallikeri in Western Lowland Gorillas and Humans
Central African Republic. Emerging Infectious Disease journal. Volume 24,
Number 8—August 2018. https://wwwnc.cdc.gov/eid/article/24/8/18-0010_article
Global Malaria Programme. A framework for malaria elimination. ISBN
978-92-4-151198-8. World Health Organization 2017, http://www.who.int/malaria/publications/atoz/9789241511988/en/
T, Kanyangarara M, Laban NM, Phiri M, Hamapumbu H, Searle KM, Stevenson JC,
Thuma PE, Moss WJ and the Southern Africa International Centers of Excellence
for Malaria Research. Characteristics of Subpatent Malaria in a Pre-Elimination
Setting in Southern Zambia. The American Journal of Tropical Medicine and
Hygiene, 10 December 2018, DOI: https://doi.org/10.4269/ajtmh.18-0399
P, Mumbengegwi D, Roberts K, Tambo M, Smith J, Whittemore B, Kelly G, Moe C,
Murphy M, Chisenga M, Greenhouse B, Ntuku H, Kleinschmidt I, Sturrock H, Uusiku
P, Gosling R, Bennett A, Hsiang MS. Subpatent malaria in a low transmission
African setting: a cross-sectional study using rapid diagnostic testing (RDT)
and loop-mediated isothermal amplification (LAMP) from Zambezi region, Namibia.
Malar J. 2018 Dec 19;17(1):480. doi: 10.1186/s12936-018-2626-5.
T, Kanyangarara M, Laban NM, Phiri M, Hamapumbu H, Searle KM, Stevenson JC,
Thuma PE, Moss WJ, For The Southern Africa International Centers Of Excellence
For Malaria Research.Characteristics of Subpatent Malaria in a Pre-Elimination
Setting in Southern Zambia. Am J Trop Med Hyg. 2018 Dec 10. doi: 10.4269/ajtmh.18-0399.
[Epub ahead of print]
[xii] Thera MA, Konea AK, Tangaraa B, Diarraa E, Niarea A, Dembeleb A, Sissokoa MS, Doumboa OK. School-aged children based seasonal malaria chemoprevention using artesunate-amodiaquine in Mali. Parasite Epidemiology and Control 3 (2018) 96–105. https://doi.org/10.1016/j.parepi.2018.02.001
Smit MR, Ochomo EO, Aljayyoussi G, Kwambai TK, Abong’o BO, Chen T, Bousema T,
Slater HC, Waterhouse D, Bayoh NM, Gimnig JE, Samuels AM, Desai MR,
Phillips-Howard PA, Kariuki SK, Wang D, Ward SA, ter Kuile FO. Safety and
mosquitocidal efficacy of high-dose ivermectin when co-administered with
dihydroartemisinin-piperaquine. www.thelancet.com/infection Published online
March 27, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30163-4